What is the difference between condition monitoring and predictive maintenance?

Condition monitoring is measuring a health signal and reacting when it changes. Predictive maintenance is using that history to estimate when a failure is likely, so work can be scheduled before it. LiveMach does the first today. We do not do the second, and we will not describe threshold alerts as prediction.

Do you predict failures?

No. Our sensors capture vibration, sound and current, and the system raises an alert when a reading crosses a threshold you and we agree. There is no model estimating remaining life or a failure date. If a vendor tells you they can predict failures on your machines from day one, ask them what baseline they used.

Do I need a vibration analyst on my team?

Not for threshold alerting. Your maintenance team needs to know what to do when an alert comes, which usually means an inspection at the next available stop. For diagnosing a difficult recurring fault on a critical machine, a trained analyst is worth the money and we will say so.

Which machines should I start with?

Start with the machine where a failure costs you the most, not the machine that breaks most often. Usually that is your bottleneck, a large motor or gearbox with a long spares lead time, or a machine that runs unattended at night. Two or three machines is a sensible first set.

Can I add this later to machines you have already fitted?

Yes. Production monitoring and condition monitoring run on the same LiveMach device and the same dashboard. Adding vibration or sound on a machine we already monitor means fitting the sensor and configuring the channel, not a fresh installation.

Will fitting sensors stop or affect my machine?

No. Sensors are fitted outside the machine, nothing inside is cut or rewired, and the device is electrically isolated. It sits alongside the machine and not in its control path, so the machine keeps running if the device fails.

What happens to the readings if the internet goes down?

The device keeps logging to its own memory and syncs on its own when the connection returns. The readings and alerts from the outage period are not lost.

What does condition monitoring cost?

It depends on how many machines you instrument and which sensors each one needs, since a vibration sensor mounted on a bearing housing is a different job from a clamp on current sensor. There is no public price list. Send us the machine list and we will quote it properly rather than giving you a number that changes at survey.

Guide

Machine condition monitoring, explained honestly

Machine condition monitoring means watching a health signal of the machine, such as vibration, sound or current, so a developing fault does not stay hidden. LiveMach fits these sensors outside the machine and raises an alert when a reading crosses an agreed threshold. It does not predict failures. It is for maintenance heads and plant owners.

A bearing does not fail all at once. It gets noisy, then it gets hot, then it seizes on a Tuesday afternoon in the middle of an order. This page explains the techniques, what they can and cannot do, and where LiveMach fits today. It is a guide, not a sales page.

Last updated: 30 September 2026

Tell us about your machines Read what we do today

How is condition monitoring different from production monitoring?

Production monitoring asks what the machine produced. Condition monitoring asks what shape the machine is in. Different question, different sensors, and usually a different person cares about the answer.

The rest of our site is about the first question. Run-time, downtime, counts, operators, cost per order. This page is about the second. Both live on the same hardware in our case, but they are genuinely separate disciplines, and confusing them is how plants end up buying something that answers a question they did not have.

Point	Production monitoring	Condition monitoring
The question	Is the machine producing, and how much?	Is the machine healthy, and is that changing?
Who asks it	Owner, plant head, production manager.	Maintenance head, reliability engineer.
Typical signal	Machine state, counts, cycle time.	Vibration, temperature, current signature, sound.
What it changes	Planning, costing, quoting, supervision.	Maintenance timing and spares planning.
Payback	Usually visible within weeks.	Depends entirely on what a failure costs you.

What are the main condition monitoring techniques?

Five techniques cover almost all condition monitoring in general manufacturing. Each one sees a different kind of fault, and none of them sees everything.

Technique	What it sees	Practical reality
Vibration analysis	Bearing wear, misalignment, imbalance, looseness, gear faults.	The most informative signal on rotating machines. Needs a sensor mounted properly on the bearing housing, and needs a baseline of that machine before a reading means anything.
Temperature	Friction, lubrication failure, overload, cooling problems.	Cheap and easy to read, but late. By the time a bearing housing is hot, the damage is usually done. Good as a backstop, weak as an early signal.
Current signature	Motor load changes, mechanical binding, phase imbalance, belt problems.	The sensor sits on the supply cable, not on the machine, so it is the easiest to fit. It sees load and electrical faults well and mechanical detail poorly.
Sound	Grinding, knocking, air leaks, cavitation.	Very good at the faults an experienced fitter already hears. Useful when the fitter is not standing there at 2 am. Noisy sheds make it harder.
Oil analysis	Wear particles, contamination, additive depletion, water ingress.	Genuinely powerful on gearboxes and hydraulics. It is a laboratory service, not a sensor, so it works on a sampling schedule. LiveMach does not do this.

LiveMach fits and reads vibration, sound and current sensors. We do not do oil analysis, and temperature only where a suitable sensor or transmitter is fitted for that deployment.

What a bearing fault looks like in a vibration signature

A healthy bearing gives a fairly even, low level vibration. A damaged one adds a repeating impulse: every time a rolling element passes over the damaged spot, it strikes, and that strike shows up as a spike at a regular interval.

As the damage grows, the spikes get larger and the intervals between them stay tied to shaft speed and bearing geometry. That regularity is the useful part. A random rise in vibration could be anything. A repeating impulse train at a rate that matches the bearing is a specific, diagnosable thing.

This is conceptual, not a lesson in vibration analysis. Reading a real signature properly is a skilled job, and on a critical machine it is worth paying somebody who does it for a living.

Healthy trace against a developing fault

The illustration below is a simulation, drawn to show the difference in shape between a healthy trace and one with a developing fault. It is not captured from a machine.

Press Simulate fault and watch the shape change. The point is the pattern, not the numbers.

Illustrative simulation

Signal within the normal band for this machine.

What does LiveMach do today?

We fit vibration and sound sensors on your machines, capture the signal continuously, set thresholds with you, and raise an alert when a reading crosses one. That is threshold based alerting, and we want to be precise about the word.

We fit the sensors
Vibration and sound sensors are mounted for the machine and the process, alongside the current sensor that already comes with power monitoring. Fitting is external, like everything else we do.
We capture continuously
The signal is sampled on the device, at a rate tuned for that machine type, so we are not flooding your network with raw waveform data.
You get an alert on a threshold
When vibration or sound crosses the level agreed for that machine, an alert goes out with the machine, the time and the reading. Your team decides what to do about it.
It joins the maintenance record
The alert and the action taken sit in the same maintenance history as your scheduled work and engineer visits, so the machine has one story rather than three.

What this does not do is tell you when the machine will fail. A threshold alert says a reading is high now. It does not forecast, it does not estimate remaining life, and we will not tell you otherwise in a demo.

Roadmap

What are we building next?

We are working on models that learn what normal looks like for your specific machine and flag a change in the signature earlier than a fixed threshold can.

The reason this is not shipped yet is not marketing caution. It is how the technique works. A model that judges whether a machine has changed needs a baseline of that machine running normally, across its real load range, its real jobs and its real seasons. Two weeks of data from a demo unit is not a baseline. That is why we start with production monitoring and threshold alerting: they are useful from day one and they build the history the models will eventually need.

When it is ready we will say so plainly, on this page, and you will be able to test it on your own machines before you pay for it.

When is condition monitoring worth it, and when is it not?

It is worth it when a failure costs far more than watching for it. It is not worth it on a machine you can replace over lunch. Sophistication is not automatically better.

Usually worth it

A machine that is the bottleneck of your whole line, where a day down is a day of orders down.
A large motor, gearbox or compressor with a long lead time on spares or a rewinding job of several days.
A machine where a failure damages the product, the die or the mould, so the repair bill is far bigger than the machine.
A machine that runs unattended at night, where a developing noise nobody hears turns into a seizure by morning.

Usually not worth it

A small motor you keep a spare of on the shelf and can swap in an hour. Run-to-failure is the correct economic decision there.
A machine that is idle half the shift anyway, where a failure costs you nothing you were going to use.
A plant that has not yet fixed the basics. If lubrication is irregular and no one records breakdown reasons, sensors will not save you.
Any machine where nobody has agreed in advance who acts on an alert. An alert nobody owns is just noise.

Our honest advice for a ten machine job shop is to start with production monitoring, get downtime and reasons under control, and add condition monitoring on the one or two machines where a failure genuinely hurts.

Which of your machines are worth instrumenting?

Send us your machine list with makes, ratings and what a day of downtime costs on each. We will tell you honestly which ones are worth instrumenting and which ones are not.

Tell us about your machines Ask us a question

Common questions

What is the difference between condition monitoring and predictive maintenance?: Condition monitoring is measuring a health signal and reacting when it changes. Predictive maintenance is using that history to estimate when a failure is likely, so work can be scheduled before it. LiveMach does the first today. We do not do the second, and we will not describe threshold alerts as prediction.
Do you predict failures?: No. Our sensors capture vibration, sound and current, and the system raises an alert when a reading crosses a threshold you and we agree. There is no model estimating remaining life or a failure date. If a vendor tells you they can predict failures on your machines from day one, ask them what baseline they used.
Do I need a vibration analyst on my team?: Not for threshold alerting. Your maintenance team needs to know what to do when an alert comes, which usually means an inspection at the next available stop. For diagnosing a difficult recurring fault on a critical machine, a trained analyst is worth the money and we will say so.
Which machines should I start with?: Start with the machine where a failure costs you the most, not the machine that breaks most often. Usually that is your bottleneck, a large motor or gearbox with a long spares lead time, or a machine that runs unattended at night. Two or three machines is a sensible first set.
Can I add this later to machines you have already fitted?: Yes. Production monitoring and condition monitoring run on the same LiveMach device and the same dashboard. Adding vibration or sound on a machine we already monitor means fitting the sensor and configuring the channel, not a fresh installation.
Will fitting sensors stop or affect my machine?: No. Sensors are fitted outside the machine, nothing inside is cut or rewired, and the device is electrically isolated. It sits alongside the machine and not in its control path, so the machine keeps running if the device fails.
What happens to the readings if the internet goes down?: The device keeps logging to its own memory and syncs on its own when the connection returns. The readings and alerts from the outage period are not lost.
What does condition monitoring cost?: It depends on how many machines you instrument and which sensors each one needs, since a vibration sensor mounted on a bearing housing is a different job from a clamp on current sensor. There is no public price list. Send us the machine list and we will quote it properly rather than giving you a number that changes at survey.

Machine downtime tracking
Breakdown reasons and maintenance history, which is where most plants should start.
OEE software
Breakdowns are one of the six big losses. This page works the whole number through.
The LiveMach sensor device
What the device reads, how it is isolated and how add on sensors connect to it.
How LiveMach pricing works
Why add ons like vibration and sound are scoped per machine, and how a pilot starts.
Talk to us about your machines
Send your machine list and we will tell you which ones are worth instrumenting.